Face identification

15
FACE FACE IDENTIFICATION IDENTIFICATION SUBMITTED BY: SUBMITTED BY: VIPIN GOYAL VIPIN GOYAL SUPERVISED BY: SUPERVISED BY: MR.SANTOSH KUMAR VERMA MR.SANTOSH KUMAR VERMA

description

Face Identification is an application that is mainly used to identify criminals based on the clues given by the eyewitnesses.

Transcript of Face identification

Page 1: Face  identification

FACE IDENTIFICATIONFACE IDENTIFICATIONSUBMITTED BY:SUBMITTED BY:

VIPIN GOYALVIPIN GOYAL

SUPERVISED BY:SUPERVISED BY:MR.SANTOSH KUMAR VERMAMR.SANTOSH KUMAR VERMA

Page 2: Face  identification

ABSTRACTABSTRACT Face Identification is an application that is mainly

used to identify criminals based on the clues given by the eyewitnesses. Based on the clues we develop an image by using the image that we have in our database and then we compare it with the images already we have. To identify any criminals we must have a record that generally contains name, age, location, previous crime, gender, photo, etc.

The primary task at hand is, given still or video images require the identification of the one or more segmented and extracted from the scene, where upon it can be identified and matched

Page 3: Face  identification

CONTENTSCONTENTS

Existing SystemExisting System Proposed SystemProposed System Architecture of the systemArchitecture of the system Algorithms proposedAlgorithms proposed Modules of the systemModules of the system UML and system designUML and system design

Page 4: Face  identification

EXISTING SYSTEMEXISTING SYSTEM The development of face identification has been past from

the year to years. In recent years to identify any criminal face they used to make a sketch or draw a image based on the eyewitnesses. It used to take more amount of time and it was very difficult task for any investigation department to easily catch the criminals within a stipulated time. In order to catch the criminals first they used to search their record whether to find out is there any record about that particular person in the past. In olden days each and every record was maintained in the books or registers or files which used to contain information about previous criminals with their names, alias name, gender, age, crime involved, etc.

Page 5: Face  identification

PROPOSED SYSTEMPROPOSED SYSTEM To overcome the drawbacks that were in the To overcome the drawbacks that were in the

existing system we develop a system that will be existing system we develop a system that will be very useful for any investigation department. very useful for any investigation department. Here the program keeps track of the record Here the program keeps track of the record number of each slice during the construction of number of each slice during the construction of identifiable human face and calculate maximum identifiable human face and calculate maximum number of slices of the similar record number. number of slices of the similar record number. Based on this record number the program Based on this record number the program retrieves the personal record of the suspect retrieves the personal record of the suspect (whose slice constituted the major parts of the (whose slice constituted the major parts of the constructed human face) on exercising the constructed human face) on exercising the “locate” option.“locate” option.

Page 6: Face  identification

ALGORITHMS PROPOSEDALGORITHMS PROPOSED

1. 1. Elastic Bunch AlgorithmElastic Bunch Algorithm – This algorithm has – This algorithm has basically three phases as follows:-basically three phases as follows:-

Firstly, we use the phase of the Firstly, we use the phase of the complex Gabor wavelet complex Gabor wavelet coefficients coefficients to achieve a more accurate location of the to achieve a more accurate location of the nodes and to disambiguate patterns which would be similar nodes and to disambiguate patterns which would be similar in their coefficient magnitudes.in their coefficient magnitudes.

Secondly, we employSecondly, we employ object adapted graphs object adapted graphs, so that , so that nodes refer to specific facial landmarks, called nodes refer to specific facial landmarks, called fiducial fiducial pointspoints. The correct correspondences between two faces . The correct correspondences between two faces can then be found across large viewpoint changes.can then be found across large viewpoint changes.

Thirdly, we have introduced a new data structure, called Thirdly, we have introduced a new data structure, called the the bunch graphbunch graph, which serves as a , which serves as a generalized generalized representation of faces representation of faces by combining jets of a small set by combining jets of a small set of individual faces. This allows the system to find the of individual faces. This allows the system to find the fiducial points in one matching process, which eliminates fiducial points in one matching process, which eliminates the need for matching each model graph individually. This the need for matching each model graph individually. This reduces computational effort significantly.reduces computational effort significantly.

Page 7: Face  identification

Eigenvalues and eigenvectorsEigenvalues and eigenvectors This algorithm is used for identifying the This algorithm is used for identifying the

correct criminal from the store database correct criminal from the store database by matching the constructed image.by matching the constructed image.

This algorithm doesn’t works well in light This algorithm doesn’t works well in light variations.variations.

It assume image as a 1D column vector It assume image as a 1D column vector with concatenated rows of pixels or 2D with concatenated rows of pixels or 2D array of pixels.array of pixels.

Then it compares the pixels of a Then it compares the pixels of a constructed image with that of prestored constructed image with that of prestored images and check out the euclidean images and check out the euclidean distance to find the best possible suspect.distance to find the best possible suspect.

Page 8: Face  identification

Pseudocode of the algorithmPseudocode of the algorithm 1. Set image resolution parameter 4 (imres)1. Set image resolution parameter 4 (imres) 2. Set PCA dimensionality parameter (PCADIM)2. Set PCA dimensionality parameter (PCADIM) 3. Read training images3. Read training images 4. Form training data matrix (Mtraindata)4. Form training data matrix (Mtraindata) 5. Form training class labels matrix (Mtrainlabels)5. Form training class labels matrix (Mtrainlabels) 6. Calculate PCA transformation matrix (tmatrix)6. Calculate PCA transformation matrix (tmatrix) 7. Calculate feature vectors of all training images 7. Calculate feature vectors of all training images

using tmatrixusing tmatrix 8. Store training feature vectors in a matrix8. Store training feature vectors in a matrix 9. Read test faces9. Read test faces 10. For each test face do10. For each test face do

Page 9: Face  identification

11. Calculate the feature vector of a test face using t matrix11. Calculate the feature vector of a test face using t matrix 12. Compute the distances between test feature vector and 12. Compute the distances between test feature vector and

all training vectorsall training vectors 13. Store the distances together with the training class 13. Store the distances together with the training class

labelslabels 14. Initialize error count to zero.14. Initialize error count to zero. 15. For each test face do15. For each test face do 16. Using the distance data, determine the person ID of the 16. Using the distance data, determine the person ID of the

most similar training vectormost similar training vector 17. If the found ID is not equal to the ID of the test image 17. If the found ID is not equal to the ID of the test image

increment error countincrement error count 18. Output the correct recognition accuracy :18. Output the correct recognition accuracy : (1 - (error count/ total test image count))*100(1 - (error count/ total test image count))*100

Page 10: Face  identification

Modules of the systemModules of the system

There are basically four modules of the There are basically four modules of the system as follows:-system as follows:-

Add ImageAdd Image - Add Image is a module that - Add Image is a module that is considered with adding image along is considered with adding image along with the complete details of the person of with the complete details of the person of whom we are taking image. whom we are taking image.

Clip Image - Clip Image - This modules main function This modules main function is to divide the images into different is to divide the images into different pieces such as hairs, forehead, eyes, nose pieces such as hairs, forehead, eyes, nose and lips and store them in the database and lips and store them in the database and also creates the files onto our system.and also creates the files onto our system.

Page 11: Face  identification

Construct Image - Construct Image - Based on the Based on the eyewitnesses we are going to construct eyewitnesses we are going to construct the images. The witness will give us the images. The witness will give us instruction by looking onto the screen on instruction by looking onto the screen on which there will be the parts of the images which there will be the parts of the images like eyes, hairs etc.like eyes, hairs etc.

Identification - Identification - This module contains the This module contains the interface to take the image from above interface to take the image from above module and it compares or searches with module and it compares or searches with the images already there in the database. the images already there in the database.

Page 12: Face  identification

Use Case DiagramUse Case Diagram

Page 13: Face  identification

Class DiagramClass Diagram

Page 14: Face  identification

Functional and Non Functional Functional and Non Functional RequirementsRequirements

Functional RequirementsFunctional Requirements:-:- The eyewitness should be able to identify The eyewitness should be able to identify

the correct image properly. There should the correct image properly. There should not be chaos for culprit’s image identity in not be chaos for culprit’s image identity in eyewitness mind.eyewitness mind.

The new database images should be The new database images should be added correctly and the culprit record added correctly and the culprit record should be updated time to time.should be updated time to time.

The system should be able to upload the The system should be able to upload the image from database correctly.image from database correctly.

The nearest match should be shown.The nearest match should be shown.

Page 15: Face  identification

Non Functional RequirementsNon Functional Requirements:-:-

Security Security - The database of criminals should be - The database of criminals should be highly secured and any modification by strangers highly secured and any modification by strangers should be strictly prohibited.should be strictly prohibited.

PerformancePerformance - The performance of the proposed - The performance of the proposed system should be accurate and fast.system should be accurate and fast.

Robust - Robust - The system should be robust and The system should be robust and vigorous.vigorous.

ReliableReliable - The system should be reliable and - The system should be reliable and should able to bear load.should able to bear load.